Coarse root biomass and architecture of hybrid aspen‘Crandon’ (Populus alba L. × P. grandidenta Michx.) grown in anagroforestry system in central Iowa, USAWilliam L. Headleea, Ronald S. Zalesny Jr.b, and Richard B. Hallc

aUA Division of Ag. Arkansas Forest Resources Center, Monticello, AR, USA; bUSDA Forest Service NorthernResearch Station, Institute for Applied Ecosystem Studies, Rhinelander, WI, USA; cDepartment of NaturalResource Ecology and Management, Iowa State University, Ames, IA, USA

ABSTRACTIn this study, we evaluated ‘Crandon’ coarse root biomass and archi-tecture grown at different topographic positions and fertilizer rates.Complete excavations were conducted on a subset of trees after thefirst growing season and showed that root biomass was stronglyrelated to stem biomass (R2 = 0.93), but not topographic positionor fertilizer rate. After the third growing season, subsamples of rootswere collected from another subset of trees and showed coarse rootarchitecture variables to be strongly related to several metrics of thetree and root size (R2 = 0.61 to 0.82), while also differing by topo-graphic position. Equations relating root biomass to stem biomasswere derived from both methodologies (complete excavation v. sub-sampling for architecture measurements), and comparison of theequations indicated no difference in slopes (p = 0.59) or intercepts(p = 0.90), although the subsampling approach had a weaker modelfit. Our results suggest ‘Crandon’ roots (i) adhere to strong allometricrelationships with stem biomass, (ii) alter their architecture within theconstraints of this allometric relationship according to site conditions,and (iii) can be subsampled to estimate root biomass from rootarchitecture parameters with similar accuracy (but less precision)compared to complete excavations.

KEYWORDSBiomass allocation; fertilizer;pipe theory; root geometry;topography

Introduction

The importance of adventitious rooting within the genus Populus has been thoroughlydescribed (Friend, Scarascia-Mugnozza, Isebrands, & Heilman, 1991; Haissig & Davis, 1994;Heilman, Ekuan, & Fogle, 1994; Pregitzer & Friend, 1996), especially with regard to theinfluence of rooting on the commercial deployment of unrooted and rooted planting stock(Zalesny, Jr., Riemenschneider, & Hall, 2005) for traditional forest products as well asproviding ecosystem services (Zalesny et al., 2016). In addition, several studies have endea-vored to quantify poplar root biomass production (Coleman, Friend, & Kern, 2004; Coyle &Coleman, 2005; DesRochers & Lieffers, 2001; Douglas, McIvor, & Lloyd-West, 2016) anddistribution in the soil profile (Douglas, McIvor, Potter, & Foote, 2010; Fortier, Truax,Gagnon, & Lambert, 2013; McIvor, Douglas, & Benavides, 2009) with different planting

CONTACT William L. Headlee headlee@uamont.edu Arkansas Forest Resources Center, UA Division of Agriculture,PO Box 3468, Monticello, AR 71656, USAColor versions of one or more of the figures in the article can be found online at www.tandfonline.com/wjsf.

JOURNAL OF SUSTAINABLE FORESTRY2019, VOL. 38, NO. 1, 18–30https://doi.org/10.1080/10549811.2018.1491861

© 2018 Taylor & Francis

materials and growing conditions, while fewer seem to have evaluated how growing condi-tions influence poplar root architecture and associated ecosystems services such as soilstabilization and erosion control (e.g., Reubens, Poesen, Danjon, Geudens, & Muys, 2007).While hybrid poplars and hybrid aspens have been primarily deployed on relatively flat orminimally sloping terrain in theMidwesternUnited States, interest in their use as componentsof agroforestry systems in the region has led to testing of their establishment potential onlandscapes with higher slopes (Headlee, Hall, & Zalesny, 2013a; Thelemann et al., 2010). Thus,the importance of understanding root system development (e.g., biomass accumulation andarchitecture) in response to site conditions such as slope and nutrient availability is vital forincreasing the success of these systems.

As a result, greater knowledge about the effects of topographic position and fertilizer oncoarse root biomass and architecture is needed. Previous research has indicated thatchanges in woody biomass allocation for poplars (Populus spp.) are often primarily afunction of tree size and genotype (Coleman et al., 2004; Coyle & Coleman, 2005; King,Pregitzer, & Zak, 1999; Wullschleger et al., 2005) rather than resource availability; in otherwords, greater resource availability simply pushes woody biomass allocation farther alonga species-specific, size-based (allometric) curve. These findings may be related to the pipetheory of plant water transport (Coutts, 1987; Shinozaki, Yoda, Hozumi, & Kira, 1964),which describes the cumulative cross-sectional area of downstream components (e.g.,stems) as being directly proportional to the cumulative cross-sectional area of upstreamcomponents (e.g., coarse roots).

Within the constraints of an allometric biomass relationship and the pipe theory,however, responses to resource availability may still occur through changes in rootarchitecture parameters such as the number, diameter, and/or length of roots (Curt,Coll, Prévosto, Balandier, & Kunstler, 2005; Di Iorio, Lassere, Scippa, & Chiatante, 2005;Mou, Mitchell, & Jones, 1997). Approaches for evaluating root architecture range fromempirical modeling (Nygren, Lu, & Ozier-Lafontaine, 2009) to highly-detailed 3D recon-struction of woody root systems (Danjon & Reubens, 2008; Danjon, Sinoquet, Godin,Colin, & Drexhage, 1999). Previous research suggests that root architecture parameters arerelatively consistent within a root system (Salas, Ozier-Lafontaine, & Nygren, 2004) andthat root biomass may be estimable from relatively few measurements of stem and rootsizes (Drexhage, Chauvière, Colin, & Nielsen, 1999). Simple empirical models based onsuch measurements would then help to alleviate the need for complete excavation of theroot system, which exhibits inherent difficulties in data collection, as well as increasedtime and labor requirements of sampling procedures relative to aboveground harvesting(Wiese, Riemenschneider, & Zalesny, 2005).

In this paper, we describe research evaluating coarse root biomass and architecture forthe hybrid aspen clone ‘Crandon’ (Populus alba L. × P. grandidenta Michx.) (Gatherum,Gordon, & Broerman, 1967; Goerndt & Mize, 2008) grown at different topographicpositions and supplied with different rates of fertilizer in central Iowa, USA. Researchon aboveground biomass at the site demonstrated that topographic position and fertiliza-tion had significant impacts on tree growth over the first three growing seasons; forexample, the highest fertilizer rate tested was associated with approximately twice asmuch aboveground biomass as the unfertilized control (Headlee et al., 2013a). To evaluatebelowground biomass, we conducted a complete excavation of root systems for a subset oftrees following the first growing season. After the third growing season, we subsampled

JOURNAL OF SUSTAINABLE FORESTRY 19

the roots systems of another subset of trees to evaluate root architecture parameters thatcould be used to estimate root system biomass. Thus, we were able to evaluate the rootbiomass production (first year) and architecture (third year) of ‘Crandon’ as functions ofallometric relationships and/or resource availability (i.e., topographic position and fertili-zer rate), and also compare root biomass equations derived from these first-year versusthird-year measurements.

Materials & methods

Site description

The study site is located adjacent to Big Creek approximately 20 km southwest of Ames,IA, and lies on an east-facing hillside which ranges in elevation from approximately 305 to325 m above sea level. Plots measuring 18.3 × 24.5 m were established at each of fivetopographic positions (floodplain, toe slope, back slope, shoulder slope, and summit), withthree plots placed along the north-south axis of each topographic position, for a total of 15plots. In the spring of 2009, two sets of trees were planted: 48 trees per plot spaced at 3.0 ×3.65 m for long-term evaluation of growth and environmental impacts relative to otherperennial and annual biomass cropping systems, and 24 trees per plot at half-spacing (1.5× 3.65 m) for short-term evaluation of the effects of topographic position and fertilizerrate during establishment (i.e., first 3 years). The short-term trees were randomly assignedto one of four fertilizer rates (0, 10, 20, or 40 g tree−1 of 20–10-5 NPK tablets; HenryField’s Seed and Nursery Co., Aurora, IN, USA) with six trees per fertilizer rate per plot.Of these, two trees per fertilizer rate per plot were randomly selected for root excavation,one after the first growing season and one after the third growing season, and are thesubject of this paper (5 topographic positions × 3 plots × 4 fertilizer rates × 2 trees = 120total trees). The soils, layout, site preparation, and planting of the tree plots have beendescribed in detail elsewhere (Headlee, 2012; Headlee et al., 2013a). Similarly, the otherbiomass cropping systems evaluated in the study have been described in detail by Ontl,Hofmockel, Cambardella, Schulte, and Kolka (2013) and Wilson et al. (2014).

Stem and root measurements

During the dormant season, stem biomass (bole and branches) of the sample trees washarvested, diameter was measured at the base of each cut stem (10 cm aboveground), thematerial was oven-dried at 100°C until stable weights were observed, and dry biomass wasmeasured to the nearest 0.1 g for each tree. The ranges of observed dry weights aresummarized in Table 1, by topographic position and fertilizer rate. For the roots, acomplete excavation was conducted with shovels for the trees harvested after the firstyear (n = 60). The excavated roots were washed, placed in paper bags, and dried andweighed in the same manner as described above for the stems. For the trees harvested afterthe third year (n = 60), a 60 cm tree spade (Caretree Systems, Inc., Columbus OH) wasused to excavate all roots within a radius of approximately 30 cm of the tree stump, andthree randomly-selected coarse roots (≥ 2 mm diameter) per tree were also excavatedusing shovels (starting at their origin from the hole created by the tree spade andextending outward through the plots). Each set of roots was placed in a plastic bag,

20 W. L. HEADLEE ET AL.

transported to the greenhouse, and kept in cold storage (5°C) until root architecturemeasurements were taken.

The coarse roots connected to the stump (n = 1,802) were individually measured atthe end severed by the tree spade for diameter (to the nearest 0.01 cm) using digitalcalipers, and the subsample of coarse roots originating from the hole created by the treespade (n = 180) were measured for length (to the nearest cm) as well as the followingdiameters: root base (where the root was severed by the tree spade), root tip (where theroot was severed during shovel excavation), and base of each root branch arising fromthe root. Individual cross-sectional area (CSA) of each root and root branch was thencalculated from these diameter measurements; total root CSA per tree and total rootbranch CSA area per root were also calculated, to facilitate simultaneous testing of thepipe theory along with root architecture. Change in root diameter and CSA (base minustip) were also calculated for each subsampled root, for use as a metric of root size thataccounts for differences in the completeness of the harvest of the root. The rootbranches were then removed and the subsampled roots were washed, placed in paperbags, and dried and weighed as previously described. The dry mass data were subse-quently used to determine root mass per unit volume, where volumes were estimatedfrom the root length and diameter measurements as described by Bolte et al. (2004). Themean ± standard error for root specific gravity was 0.345 ± 0.008 dry g cm−3, which isslightly lower than the value of around 0.39 g cm−3 typically observed for the stem woodof ‘Crandon’ trees (Hall, Hart, & Peszlen, 2001; Headlee et al., 2013b).

Estimation of total coarse root biomass

Root architecture variables were used to estimate total coarse root biomass with the assump-tions that 1) the architecture of coarse roots can be reasonably described as a series ofconnected cylinders which reduce in size each time a root branch diverges from the root, 2)root branches, and any subordinate branches arising from them, will have similar architectureas the roots and root branches from which they arise, and 3) coarse roots and their rootbranches adhere to the pipe theory of plant water transport. A schematic of the simplified rootstructure associated with these assumptions is shown in Figure 1. The first assumption allows

Table 1. Summary of stem biomass and root dry biomass data for 1-year-old and 3-year-old trees, bytopographic position and fertilizer rate. The data for 3-year-old root biomass are estimates based onroot architecture; all other data are measured values.

1-Year-Old Trees 3-Year-Old Trees

Stem Biomass (g) Root Biomass (g) Stem Biomass (g) Root Biomass (g)

Topographic PositionSummit 7.7–69.4 7.9–52.2 170–1972 61–730Shoulder 1.6–75.8 3.8–60.0 613–1994 362–963Backslope 3.9–98.7 7.5–59.2 472–3490 157–1748Toeslope 1.6–99.7 4.0–81.9 184–2746 182–1216Floodplain 2.1–55.6 4.6–65.1 447–3958 55–1799

Fertilizer (g tree−1)0 2.1–45.5 4.6–48.9 472–1910 79–60110 3.6–74.9 7.5–81.9 205–2024 55–80920 1.6–90.9 3.8–64.7 170–2624 143–86340 1.6–99.7 4.0–65.1 428–3958 157–1799

JOURNAL OF SUSTAINABLE FORESTRY 21

the volume of each root to be calculated from the lengths and CSA of its cylinders. The secondassumption is based on previous research with fractal root models suggesting that rootarchitecture parameters are similar across root orders (Salas et al., 2004) and allows rootarchitecture parameters to be estimated from primary roots alone, which greatly reduces thetime and labor required for excavation and measurements. The third assumption allows thesize and number of roots to be estimated from the size of the stem, and the size and number ofroot branches to be estimated from the size of the root. Together, these assumptions allow forestimation of the volume of a root as

VR¼X

ðAi�LiÞ¼A0�30þA1�L1þA2�L2 (1)

where VR is estimated volume of the root (cm3), Ai is the area of the ith cylinder along theroot (cm2), and Li is the length of the ith cylinder along the root (cm). A length of 30 cm isshown here for the length of the root segment attached to the stump (A0) as this was theradius of the hole excavated by the tree spade, because root branches lack these segmentsattached to the stump, this term is excluded when estimating root branch volumes.

Previous research indicates that the use of the means of root architecture parametersproduce similar results as more detailed methods while allowing much simpler calcula-tions (Nygren et al., 2009); thus, root volumes in this study were estimated using meanCSA (e.g., total CSA of roots divided by number of roots) rather than estimated for eachindividual root or root branch. Similarly, the length between each root branch and thereduction in cylinder area at each root branch were also treated as equal to their means

Figure 1. Diagram of the simplified architecture of ‘Crandon’ roots assumed for the current study. Rootsand root branches are assumed to be analogous to a series of cylinders and adhere to the pipe theoryof plant water transport.

22 W. L. HEADLEE ET AL.

(e.g., the total change in root CSA divided by the number of root branches). Based onthese volume estimates, total root biomass was then estimated as

BR¼ 0:345 � NR VR þ NX1 VX1 þ NX2 VX2. . .ð Þð Þð Þ (2)

where BR is estimated total root biomass (g), 0.345 is the observed mean root weight perunit volume (g cm−3; described in the preceding section), NR is the number of roots, VR isthe volume per root (cm3), NXi is the number of ith-order root branches, and VXi is thevolume per branch of ith-order root branches (cm3). Because the estimated number ofthird-order root branches per second-order root branch was found to be less than one(mean = 0.8 branches), we concluded our calculations of total root volume and biomasswith second-order root branches.

Data analysis

The root biomass data for the 1-year-old trees were analyzed as a two-way factorial of topo-graphic position and fertilizer rate (fixed effects, completely randomized design), with stembiomass used as the covariate and random effects of plot and tree within the plot. Both rootbiomass and stem biomass were log-transformed to homogenize variance. For the 3-year-oldtrees, the root architecture variables of total CSA of the roots (covariate = CSA of the stem) andnumber of roots (covariate = stem diameter) were also analyzed as a completely randomized,fixed effects, two-way factorial of topographic position and fertilizer rate, with random effects ofplot and tree within plot. The sub-sampling of roots allowed the remaining variables (rootlength, total CSA of root branches, and number of root branches) to be analyzed with theadditional random effect of root within the tree, and the variability among trees to be testedagainst the variability among roots within trees. Number of root branches was used as thecovariate for root length, change in CSA of the root was used as the covariate for total CSA of theroot branches, and change in diameter of the root was used as the covariate for the number ofroot branches. Analysis of variance (ANOVA) was conducted using PROC MIXED in SAS®(SAS Institute Inc., Cary, NC, USA), with denominator degrees of freedom determined via theKenward-Rogersmethod (Littell, Stroup, & Freund, 2002). Based on the results of the ANOVAs,model factors determined to be statistically significant (p < 0.05) were used to develop regressionequations with PROC GLM in SAS®. Statistical contrasts in PROC GLM were also used toevaluate the biomass equations developed from the first-year and third-year measurements fordifferences in slopes and intercepts (Littell et al., 2002).

Results

For trees harvested after the first growing season, root biomass was found to be signifi-cantly influenced by stem biomass (p < 0.0001), but not by topographic position orfertilizer rate (Table 2). Additionally, the relationship between root biomass and stembiomass was found to be very strong (R2 = 0.93). Thus, coarse root biomass appeared to beprimarily a function of stem size, rather than resource availability as influenced bytopographic position and fertilizer rate.

For trees harvested after the third growing season, the analyses indicated that thetotal CSA of roots was significantly influenced by (p < 0.0001) and strongly related to

JOURNAL OF SUSTAINABLE FORESTRY 23

(R2 = 0.82) the CSA of the stem but was not significantly influenced by topographicposition or fertilizer rate. This indicates that the roots adhered to the pipe theory,regardless of resource availability. The number of roots was significantly related tostem diameter (p < 0.0001) as well as topographic position (p = 0.04), indicating that‘Crandon’ does alter this aspect of its rooting in response to site conditions.Comparison of means by topographic position (Figure 2(a)) showed that the floodplainhad significantly more roots per unit stem diameter (14.5 roots per cm) than the othertopographic positions (8.9 to 11.3 roots per cm). Linear regression using these

Table 2. Results of ANOVA and regression analyses for ‘Crandon’ root biomass (BR; g), sum of root cross-sectional area (∑AR; cm

2), number of coarse roots (NR), root length (LR; cm), sum of root branch cross-sectional area (∑AX; cm

2), and number of root branches (NX). Covariates for each parameter are shownin the Equation column and include stem biomass (BS; g), cross-sectional area of the stem (AS; cm

2),diameter of the stem (DS; cm), number of root branches (NX), change in root cross-sectional area (ΔAR;cm2), and change in root diameter (ΔDR; cm). Intercepts were suppressed when not significantlydifferent from zero (∑AR, NR, ∑AX, and NX), and treatment-specific coefficients (a, b, and c) were fitwhen significant treatment effects were observed (NR, LR, and NX).

ANOVA (p-values) Regression

Parameter Covariate Position Fertilizer Pos. × Fert. Tree Equation R2

1-Year-Old TreesBR < 0.0001 0.77 0.42 0.39 n/a = 2.66BS°

.68 0.933-Year-Old Trees∑AR < 0.0001 0.42 0.16 0.26 n/a = 1.35AS 0.82NR < 0.0001 0.04 0.66 0.60 n/a = aDS 0.73LR < 0.0001 0.01 0.83 0.94 0.03 = b + 13.36NX 0.61∑AX < 0.0001 0.72 0.22 0.46 0.23 = 0.475ΔAR 0.70NX < 0.0001 0.87 0.86 0.82 0.03 = cΔDR 0.81

Figure 2. Comparisons among topographic positions for coefficients a (number of roots per unit stemdiameter; panel A) and b (root length intercept; panel B). Statistically significant differences (p < 0.05)are indicated by different letters above the standard error bars.

24 W. L. HEADLEE ET AL.

position-specific coefficients demonstrated that the number of roots had a relativelystrong relationship with stem diameter (R2 = 0.73).

Root length was found to be significantly related to the number of root branches (p <0.0001) and topographic position (p = 0.01), and also differed significantly between treeswithin treatments (p = 0.03); this suggests that ‘Crandon’ alters its root length in response toboth broad scale (topographic) and micro-site (between-tree) conditions. Comparison ofmeans by topographic position (Figure 2(b)) showed that the floodplain had a significantlylower root length intercept (55 cm) than the other topographic positions (129 to 158 cm).Because the tree effects also capture topographic position effects, tree-specific coefficients werefit in the regression analysis and resulted in a moderately strong relationship between rootlength and the number of root branches (R2 = 0.61). Similar to the total CSA of the roots, thetotal CSA of the root branches was significantly related to only the covariate (change in CSA ofthe root; p < 0.0001), with a relatively strong model fit (R2 = 0.70). This indicates that the rootbranches also adhered to the pipe theory, regardless of resource availability. The number ofroot branches was significantly related to the change in diameter of the root (p < 0.0001) andalso differed significantly between trees within treatments (p = 0.03), indicating that ‘Crandon’alters its root branching in response tomicro-site conditions. The regression analysis includedtree-specific coefficients for the number of root branches per unit change in root diameter andshowed a strong model fit (R2 = 0.81).

The estimates of 3-year-old coarse root biomass derived from the root architectureparameters and dry weight data had a moderately strong relationship with stem biomass(R2 = 0.56) and produced a similar equation as was observed for the 1-year-old trees(Figure 3). Statistical contrasts revealed no significant differences between the two equationsin terms of slope (p = 0.59) or intercept (p = 0.90). Thus, subsampling of 3-year-old treesproduced a similar equation for predicting root biomass from stem biomass (albeit with aweaker fit) compared to complete excavations of 1-year-old trees.

Discussion

In terms of our underlying assumptions, the similarity between the root biomass equationdeveloped from complete excavations and that developed from our root architecturemodel suggests that ‘Crandon’ root architecture can be reasonably approximated as aseries of connected cylinders (first assumption). The ANOVA results also suggest a certainamount of consistency in root architecture within a given root system (second assump-tion), in that the variability among trees was typically greater than the variability amongroots within trees. We also specifically tested and confirmed that the roots and their rootbranches conform to the pipe theory (third assumption), as the sum of the CSA of theroots and root branches were strongly related to the CSA of the stem and change in CSAof the root, respectively. Thus, the underlying assumptions of the model appear to bereasonable, to the extent that we were able to test them in the current study. While ourroot biomass estimates for the 3-year-old trees included extrapolation to second-orderroot branches, it should be noted that the contribution of these roots to the estimates oftotal root biomass was relatively small (mean = 11% of total estimated root biomass)compared to roots (mean = 67%) and first-order root branches (mean = 22%).

While research on stem biomass at our site showed significant differences due totopographic position and fertilization, the same research also showed that allocation of

JOURNAL OF SUSTAINABLE FORESTRY 25

stem biomass between boles and branches was primarily a function of tree size (Headleeet al., 2013a). Similarly, the current study shows that root biomass was primarily afunction of tree size for 1-year-old trees, and root biomass estimates derived from rootarchitecture relationships suggest a comparable (albeit weaker) trend for 3-year-old trees.This weaker model fit may be at least partially attributable to the use of estimated meanroot architecture parameters as well as the nested structure of Equation 2, which amplifiesthe errors associated with the estimated means. Of particular interest is the potential errorintroduced through the root length parameter, as it had the weakest model fit (R2 = 0.61)of those evaluated in the current study. Thus, additional work aimed at improving rootlength estimates appears warranted. For example, greater predictive power may be attain-able by subsampling a larger number of roots per tree, as our sub-sampling of 3 roots pertree was on average only about 10% of the total number of roots.

Similarly, model fit for root length and possibly other root architecture variables mightalso be improved by accounting for differences in growing conditions at the micro-sitelevel. Previous research has shown that variability among (and even within) individual treeroots can occur when heterogeneous soil conditions such as patches of nutrient-rich soilare present, particularly with hardwood trees (Mou et al., 1997). Our results suggest thatthe majority of the variation in ‘Crandon’ root architecture was explained by allometricrelationships along with topographic and between-tree differences, as indicated by the R2

values in Table 2. However, it is possible that much of the remaining variability representsindividual roots (and/or root segments) deviating about the means of the root architectureparameters. Thus, information on micro-site conditions may be useful for improvingmodel fit, although incorporation of such information would likely require considerablymore complex models.

Figure 3. Root biomass measured after the first growing season (circles) and estimated via rootarchitecture parameters after the third growing season (squares) for ‘Crandon’ trees, in relation tomeasured stem biomass.

26 W. L. HEADLEE ET AL.

Regarding topographic effects on root architecture, the floodplain was observed to havemore roots per unit stem diameter and less length per root compared to the othertopographic positions (Figure 2). Because the sum of the CSA of the roots is proportionalto stem area, as dictated by the pipe theory, the greater number of roots per unit stem sizealso implies a smaller average CSA per root. Thus, for a given stem diameter, the flood-plain trees had a larger number of roots that were shorter in length and smaller indiameter than the other topographic positions. Brassard et al. (2009) describe a tendencyfor trees to produce more numerous roots concentrated closer to the tree under condi-tions of higher resource availability; this may explain the changes in root architecture forthe floodplain in the current study, as research with the other cropping systems at our siteshowed the floodplain to have higher levels of soil nitrogen, organic matter, and waterconductivity than the other topographic positions (Ontl et al., 2013). The inverse nature ofthe relationship between the number and size of roots also indicates a trade-off in rootarchitecture that allows the trees to respond to site conditions while maintaining anallometric relationship between aboveground and belowground biomass.

While the results of the current study are encouraging, we recommend broader testingof these methods using additional genotypes and/or sites, especially given the potential forplasticity in rooting across variable environments (Pregitzer & Friend., 1996). Detailedinformation about the orientation of the roots (e.g., uphill v. downhill direction, horizon-tal v. diagonal v. vertical angles) could also be included in future studies to produce spatialmodels of root systems. We collected basic spatial orientation data on a subset of our 3-year-old trees, and the results (though insufficiently replicated for statistical analyses)indicate that the percentages of roots oriented uphill and downhill were approximately50% each, and the percentages oriented horizontally (roughly < 30 degrees relative toground level), diagonally (roughly 30 to 60 degrees), and vertically (roughly > 60 degrees)were approximately 65%, 20%, and 15%, respectively. These observations were similaracross topographic positions, which is in contrast to reports of differences in spatialdistribution of roots associated with topography (Di Iorio et al., 2005; McIvor et al.,2009); however the aforementioned studies involved older trees (i.e., ≥ 10 years old), andtherefore may reflect an age effect that had yet to develop at our site. The observation thatthe majority of roots were oriented horizontally is consistent with previous descriptions ofpoplar roots being mostly horizontal and near the soil surface (Pregitzer & Friend, 1996).

Conclusion

In summary, our results demonstrate that coarse root biomass in hybrid aspen ‘Crandon’is primarily a function of tree size, that ‘Crandon’ alters its root architecture in response tosite conditions within the constraints of this allometric relationship, and that a comparableroot biomass equation may be developed using root architecture information from sub-sampling techniques versus complete excavations of root systems. Among the rootarchitecture variables evaluated here, root length appears especially deserving of additionalstudy aimed at improving model fit. We thus encourage wider testing with ‘Crandon’ aswell as other clones and species, particularly to determine the extent to which greatersubsampling and/or other adjustments may improve the fit of biomass equations derivedfrom root architecture measurements.

JOURNAL OF SUSTAINABLE FORESTRY 27

Acknowledgments

We would like to dedicate this paper to the memory of our dear friend and co-author Rick Hall,who passed away in September 2016. We would also like to thank ArborGen, Inc. (Ridgeville SC),for supplying a portion of the long-term study trees, ISU Horticulture Research Station for the useof their tree spade, and Ed Bauer and Austin Himes for providing helpful comments on an earlierdraft of the manuscript.

Funding

Funding and other support for this research was provided by the Leopold Center for SustainableAgriculture (Project E2008-24), USDA Agriculture and Food Research Initiative (ProjectIOW5249), USDA National Institute of Food and Agriculture (McIntire Stennis Project 1009221),Iowa State University College of Agriculture and Life Sciences, and USDA Forest Service NorthernResearch Station Institute for Applied Ecosystem Studies.

References

Bolte, A., Rahmann, T., Kuhr, M., Pogoda, P., Murach, D., & Gadow, K. V. (2004). Relationshipsbetween tree dimension and coarse root biomass in mixed stands of European beech (Fagussylvatica L.) and Norway spruce (Picea abies [L.] Karst.). Plant and Soil, 264, 1–11. doi:10.1023/B:PLSO.0000047777.23344.a3

Brassard, B. W., Chen, H. Y. H., & Bergeron, Y. (2009). Influence of environmental variability onroot dynamics in northern forests. Critical Reviews in Plant Science, 28, 179–197. doi:10.1080/07352680902776572

Coleman, M. D., Friend, A. L., & Kern, C. C. (2004). Carbon allocation and nitrogen acquisition ina developing Populus deltoides plantation. Tree Physiology, 24, 1347–1357.

Coutts, M. P. (1987). Developmental processes in tree root systems. Canadian Journal of ForestResearch, 17, 761–767. doi:10.1139/x87-122

Coyle, D. R., & Coleman, M. D. (2005). Forest production responses to irrigation and fertilizationare not explained by shifts in allocation. Forest Ecology and Management, 208, 137–152.doi:10.1016/j.foreco.2004.11.022

Curt, T., Coll, L., Prévosto, B., Balandier, P., & Kunstler, G. (2005). Plasticity in growth, biomassallocation and root morphology in beech seedlings as induced by irradiance and herbaceouscompetition. Annals of Forest Science, 62, 51–60. doi:10.1051/forest:2004092

Danjon, F., & Reubens, B. (2008). Assessing and analyzing 3D architecture of woody root systems, areview of methods and applications in tree and soil stability, resource acquisition and allocation.Plant and Soil, 303, 1–34. doi:10.1007/s11104-007-9470-7

Danjon, F., Sinoquet, H., Godin, C., Colin, F., & Drexhage, M. (1999). Characterisation of structuraltree root architecture using 3D digitizing and AMAPmod software. Plant and Soil, 211, 241–258.doi:10.1023/A:1004680824612

DesRochers, A., & Lieffers, V. J. (2001). Root biomass of regenerating aspen (Populus tremuloides)stands of different densities in Alberta. Canadian Journal of Forest Research, 31, 1012–1018.doi:10.1139/x01-037

Di Iorio, A., Lassere, B., Scippa, G. S., & Chiatante, D. (2005). Root system architecture of Quercuspubescens trees growing on different sloping conditions. Annals of Botany, 95, 351–361.doi:10.1093/aob/mci033

Douglas, G. B., McIvor, I. R., & Lloyd-West, C. M. (2016). Early root development of field-grownpoplar: Effects of planting material and genotype. New Zealand Journal of Forestry Science, 46, 1.doi:10.1186/s40490-015-0057-4

28 W. L. HEADLEE ET AL.

Douglas, G. B., McIvor, I. R., Potter, J. F., & Foote, L. G. (2010). Root distribution of poplar atvarying densities on pastoral hill country. Plant and Soil, 333, 147–161. doi:10.1007/s11104-010-0331-4

Drexhage, M., Chauvière, M., Colin, F., & Nielsen, C. N. N. (1999). Development of structural rootarchitecture and allometry of Quercus petraea. Canadian Journal of Forest Research, 29, 600–608.doi:10.1139/x99-027

Fortier, J., Truax, B., Gagnon, D., & Lambert, F. (2013). Root biomass and soil carbon distributionin hybrid poplar riparian buffers, herbaceous riparian buffers and natural riparian woodlots onfarmland. SpringerPlus, 2, 539. doi:10.1186/2193-1801-2-539

Friend, A. L., Scarascia-Mugnozza, G., Isebrands, J. G., & Heilman, P. E. (1991). Quantification oftwo-year-old hybrid poplar root systems: Morphology, biomass, and 14C distribution. TreePhysiology, 8, 109–119.

Gatherum, G. E., Gordon, J. C., & Broerman, B. F. S. (1967). Effects of clone and light intensityon photosynthesis, respiration and growth of aspen-poplar hybrids. Silvae Genetica, 16(4),128–132.

Goerndt, M. E., & Mize, C. (2008). Short-rotation woody biomass as a crop on marginal lands inIowa. Northern Journal of Applied Forestry, 25, 82–86.

Haissig, B. E., & Davis, T. D. (1994). A historical evaluation of adventitious rooting research to1993. In T. D. Davis & B. E. Haissig (Eds.), Biology of adventitious root formation (pp. 275–331).New York, NY: Plenum Press.

Hall, R. B., Hart, E. R., & Peszlen, I. (2001). Modifying Populus environmental responses: Impactson wood quantity and quality. In N. Morohoshi & A. Komamine (Eds.), Molecular breeding ofwoody plants (pp. 229–238). Amsterdam, the Netherlands: Elsevier Science B.V.

Headlee, W. L. (2012). Productivity and biometry of hybrid poplars with respect to establishment,regeneration, regional modeling, and utilization of bio-energy byproducts. PhD dissertation, IowaState University, Ames, IA. 140 p.

Headlee, W. L., Hall, R. B., & Zalesny, R. S. (2013a). Establishment of alleycropped hybrid aspen“Crandon” in central Iowa, USA: Effects of topographic position and fertilizer rate on above-ground biomass production and allocation. Sustainability, 5, 2874–2886. doi:10.3390/su5072874

Headlee, W. L., Zalesny, R. S., Hall, R. B., Bauer, E. O., Bender, B., Birr, B. A., . . . Wiese, A. H.(2013b). Specific gravity of hybrid poplars in the north-central region, USA: Within-tree varia-bility and site × genotype effects. Forests, 4, 251–269. doi:10.3390/f4020251

Heilman, P. E., Ekuan, G., & Fogle, D. B. (1994). First-order root development from cuttings ofPopulus trichocarpa × Populus deltoides hybrids. Tree Physiology, 14, 911–920.

King, J. S., Pregitzer, K. S., & Zak, D. R. (1999). Clonal variation in above- and below-groundgrowth responses of Populus tremuloides Michaux: Influence of soil warming and nutrientavailability. Plant and Soil, 217, 119–130. doi:10.1023/A:1004560311563

Littell, R. C., Stroup, W. W., & Freund, R. J. (2002). SAS® for linear models (4th ed., pp. 466). Cary,NC: SAS Institute Inc.

McIvor, I. R., Douglas, G. B., & Benavides, R. (2009). Coarse root growth of Veronese poplar treesvaries with position on an erodible slope in New Zealand. Agroforestry Systems, 76, 251–264.doi:10.1007/s10457-009-9209-y

Mou, P., Mitchell, R. J., & Jones, R. H. (1997). Root distribution of two tree species under aheterogeneous nutrient environment. Journal of Applied Ecology, 34, 645–656. doi:10.2307/2404913

Nygren, P., Lu, M., & Ozier-Lafontaine, H. (2009). Effects of turnover and internal variability of treeroot systems on modeling coarse root architecture: Comparing simulations for young Populusdeltoides with field data. Canadian Journal of Forest Research, 39, 97–108. doi:10.1139/X08-158

Ontl, T. A., Hofmockel, K. S., Cambardella, C. A., Schulte, L. A., & Kolka, R. K. (2013).Topographic and soil influences on root productivity of three bioenergy cropping systems.New Phytologist, 199, 727–737. doi:10.1111/nph.12302

Pregitzer, K. S., & Friend, A. L. (1996). The structure and function of Populus root systems. In R. F.Stettler, H. D. Bradshaw Jr., P. E. Heilman, & T. M. Hinckley (Eds.), Biology of populus and itsimplications for management and conservation (pp. 331–354). Ottawa, Canada: NRC Res Press.

JOURNAL OF SUSTAINABLE FORESTRY 29

Reubens, B., Poesen, J., Danjon, F., Geudens, G., & Muys, B. (2007). The role of fine and coarseroots in shallow slope stability and soil erosion control with a focus on root system architecture:A review. Trees, 21, 385–402. doi:10.1007/s00468-007-0132-4

Salas, E., Ozier-Lafontaine, H., & Nygren, P. (2004). A fractal root model applied for estimating theroot biomass and architecture in two tropical legume tree species. Annals of Forest Science, 61,337–345. doi:10.1051/forest:2004027

Shinozaki, K., Yoda, K., Hozumi, K., & Kira, T. (1964). A quantitative analysis of plant form – Thepipe model theory. I. Basic analysis. Japanese Journal of Ecology, 14, 97–105.

Thelemann, R., Johnson, G., Sheaffer, C., Banerjee, S., Cai, H., & Wyse, D. (2010). The effect oflandscape position on biomass crop yield. Agronomy Journal, 102, 513–522. doi:10.2134/agronj2009.0058

Wiese, A. H., Riemenschneider, D. E., & Zalesny, R. S., Jr. (2005). An inexpensive rhizotron designfor two-dimensional, horizontal root growth measurements. Tree Planters’ Notes, 51, 40–46.

Wilson, D. M., Heaton, E. A., Schulte, L. A., Gunther, T. P., Hall, R. B., Headlee, W. L., . . . Boersma,N. N. (2014). Establishment and short-term productivity of annual and perennial bioenergy cropsacross a landscape gradient. BioEnergy Research, 7, 885–898. doi:10.1007/s12155-014-9409-9

Wullschleger, S. D., Yin, T. M., DiFazio, S. P., Tschaplinski, T. J., Gunter, L. E., Davis, M. F., &Tuskan, G. A. (2005). Phenotypic variation in growth and biomass distribution for twoadvanced-generation pedigrees of hybrid poplar. Canadian Journal of Forest Research, 35,1779–1789. doi:10.1139/x05-101

Zalesny, R. S., Jr., Stanturf, J. A., Gardiner, E. S., Perdue, J. H., Young, T. M., Coyle, D. R., . . . Hass,A. (2016). Ecosystem services of woody crop production systems. BioEnergy Research, 9(2), 465–491. doi:10.1007/s12155-016-9737-z

Zalesny, R. S., Jr., Riemenschneider, D. E., & Hall, R. B. (2005). Early rooting of dormant hardwoodcuttings of Populus: Analysis of quantitative genetics and genotype × environment interactions.Canadian Journal of Forest Research, 35, 918–929. doi:10.1139/x05-018

30 W. L. HEADLEE ET AL.